Citations

Photoplethysmography (PPG) is a non-invasive optical technique commonly used to measure heart rate and oxygen saturation, but its waveform contains additional physiological information that can be analyzed using artificial intelligence (AI). This narrative review summarizes the emerging applications of AI-based PPG in cardiovascular, respiratory, sleep, hemodynamic, pregnancy-related, and portal-hypertension assessment, with the aim of evaluating its potential beyond conventional monitoring and identifying barriers to clinical translation. The literature search was conducted using PubMed, Google Scholar, IEEE Xplore, ScienceDirect, and SpringerLink. Additional relevant studies were identified through screening the reference lists of included articles. Studies published between 2002 and 2026 were identified to capture the development of PPG from conventional monitoring to newer AI-based applications. Human studies were prioritized, while relevant computational, simulated, synthetic, ex vivo, and technical studies were also included. Studies unrelated to PPG, duplicates, and studies with limited relevance were excluded. A total of 96 references were included, covering AI approaches such as convolutional and deep neural networks, ensemble methods, transfer learning, U-Net, generative adversarial networks, and Transformer-based models. Overall, the reviewed evidence suggests that AI-based PPG may support blood pressure estimation, atrial fibrillation detection, sleep and respiratory monitoring, vascular aging assessment, pulmonary hypertension screening, preeclampsia assessment, volume-status and compensatory-reserve assessment, and exploratory assessment related to portal hypertension. However, clinical translation remains limited by motion artifacts, sensor and measurement-site variability, skin-pigmentation-related bias, physiological and environmental influences, heterogeneous methods, limited external validation, and inconsistent clinical and regulatory standards.

Artificial intelligence (AI) is rapidly transforming healthcare through applications in clinical decision support, diagnostic imaging, population health management, and workflow optimization. Despite these advances, real-world deployment continues to expose critical challenges related to safety, bias, transparency, and integration into clinical workflows. Algorithmic bias can exacerbate health disparities, limited explainability may undermine clinician trust, and insufficient validation and post-deployment monitoring can compromise patient safety. Although the World Health Organization (WHO) has established six ethical principles for AI in health, including autonomy, well-being and safety, transparency, accountability, equity, and sustainability, translating these high-level principles into practical and enforceable governance mechanisms remains a persistent challenge. This narrative review synthesizes insights from bioethics, health policy, computer science, and clinical medicine to identify gaps in current AI governance approaches and proposes a lifecycle-aligned governance-by-design framework that operationalizes WHO ethical principles across key stages of the healthcare AI lifecycle, including data collection, model development, validation, deployment, and post-deployment monitoring. The framework integrates concrete governance mechanisms such as consent governance, fairness evaluation, external validation, explainability, clinician oversight, and continuous performance monitoring. Overall, this work advances a practical, lifecycle-integrated approach to AI governance and provides a structured foundation for developing safe, equitable, and trustworthy AI systems in healthcare.

Abstract Multiple myeloma (MM) with t(11;14)(CCND1;IGH) remains the only subset sensitive to the BCL2 inhibitor venetoclax. However, not all patients with t(11;14)(CCND1;IGH) respond to treatment, and some progress early after initial response. To investigate this, we examined 44 whole-genome and whole-exome sequencing data samples from 34 patients with t(11;14) MM treated with venetoclax. The presence of mutations in the RAS pathway was strongly associated with shortened progression-free survival (PFS) and was validated in an independent cohort of 21 patients with MM. The presence of 1q gain was also associated with shorter PFS in patients without RAS mutations. In 10 patients with paired prevenetoclax and postvenetoclax treatment samples, postvenetoclax progression was recurrently driven by the selection of genomic events in the BCL2/MCL1 and RAS pathways and of high-risk features (eg, loss of TP53 and CDKN2C). Overall, our study shows that comprehensive genomic profiling can identify most mechanisms underlying resistance to BCL2 inhibition in t(11;14)(CCND1;IGH) MM.

Abstract Background: Sickle cell disease (SCD) is marked by wide variability in clinical phenotypes and severity due to environmental, clinical, and genetic factors. Variants in iron-regulatory genes are well described modifiers of clinical presentation in different diseases, but their role in modulating SCD phenotype has been sparsely studied in African populations. In this cross-sectional study, we investigated the role of single nucleotide polymorphisms (SNPs) in iron-regulatory genes in SCD severity and lifetime transfusion(Tx) requirement. Methods: The study was approved by the institutional review board of Lagos University Teaching Hospital, Lagos, Nigeria. Clinical data were abstracted from medical records and patient interviews using a structured questionnaire. Participants were recruited in steady state and were stratified according to the severity score previously described by Shah et al. (Clinicoecon Outcomes Res. 2020;12:625). Biochemical analyses included serum transferrin and serum iron; transferrin saturation (TS) was calculated. Genotyping was performed using the Axiom Precision Medicine Research Array (PMRA; Thermo Fisher Scientific). Seventy-one single nucleotide polymorphisms (SNPs) in genes implicated in iron metabolism were analyzed, including SLC40A1 (ferroportin), HFE (homeostatic iron regulator), HJV (hemojuvelin), HAMP (hepcidin), and TFR2 (transferrin receptor 2). SNPs whose variants were not represented in this cohort, including HFE C282Y and H63D, were excluded. Ensembl and NCBI databases were used to describe the wild-type (WT)/variant alleles and gene consequences. Univariate and multivariate logistic regression were used to analyze binary outcomes, and age-adjusted linear regression for Tx burden. Odds ratios (OR) and regression coefficients (β) were estimated using the WT as the reference. Multiple-testing correction was performed as necessary. Statistical analyses were conducted in R version 4.4.1, with p<0.05 considered significant. Results: We included 60 genotyped, confirmed HbSS patients; median age was 22 years (range: 18–43), 22 (37.3%) were female, 42 (70.0%) were classified as mild severity (Class I), 5 (8.0%) as Class II, and 13 (22.0%) as severe (Class III). Twenty-seven (45.0%) had never received a blood Tx, while 33 (55.0%) had received Tx, median of 2units (range: 1-10). On univariate analysis (UVA), five SNPs (rs10218795, rs10188230, rs114250626, rs4287798, and rs634349) were significantly associated with severe phenotype; three of these remained significant on multivariable analysis (MVA). Two were associated with higher odds of severe phenotype: rs10177654 (OR = 28.9, 95% CI: 3.0–100.0, p = 0.03) and rs10218795 (OR = 15.0, 95% CI: 2.4–327.0, p<0.01), while rs10188230 was associated with lower odds (OR = 0.03, 95% CI: 0.008–0.28, p<0.01). Results remained consistent after adjusting for age. For lifetime transfusion burden, both UVA and MVA identified two significant SNPs: rs4667289, associated with higher transfusion burden (β = 2.6, 95% CI: 0.04–5.2, p = 0.047) and rs11568344, associated with lower transfusion burden (β = –1.8, 95% CI: –3.5 to –0.36, p = 0.02). Genotype distributions for the five significant SNPs were as follows: rs10177654 (SLC40A1, transcription factor binding site), TT(WT) 12 (20.3%), TC 30 (50.8%), CC 17 (28.8%); rs10188230 (SLC40A1, intron variant), TT (WT) 32 (54.2%), TC 22 (37.3%), CC 5 (8.5%); rs10218795 (HJV, intron variant), CC (WT) 22 (37.3%), TC 29 (49.2%), TT 8 (13.6%); rs4667289 (SLC40A1 proximal intergenic), GG (WT) 46 (78.0%), AG 10 (16.9%), AA 3 (5.1%); rs11568344 (SLC40A1 splice region variant), GG (WT) 47 (79.7%), AG 9 (15.3%), AA 3 (5.1%). Compared with the homozygous WT, patients with rs10188230 had a lower median TS (21.65% vs. 43.3%, p = 0.04), while those with rs4667289 had a higher median TS (83% vs. 31%, p<0.01). When adjusted for age and lifetime Tx, rs4667289 remained significant (β = 47.8, 95% CI 3.6-82.8, p = 0.03). No other SNPs showed association with TS. Conclusion: We present novel genetic data on iron-regulatory variants in African patients with sickle cell disease, identifying SNPs in SLC40A1 and HJV associated with disease severity. Most of these are yet to be fully described in the literature, and their intronic locations suggest possible regulatory roles. We hope to further validate these findings in larger cohorts and through functional studies to clarify their impact on iron metabolism and clinical significance.

Abstract Cyclin D1, encoded by CCND1, is a cell cycle regulator, overexpressed in ~15% of multiple myeloma (MM) cases via CCND1::IGH super-enhancer juxtaposition through t(11;14). Cyclin D1 also plays a non-canonical role in transcriptional regulation, but in t(11;14) MM its function in gene regulation remains undefined. We investigated the global transcriptional and epigenomic consequences of CCND1overexpression in t(11;14) MM. To define a CCND1 signature, RNA-seq from two independent primary MM patient cohorts, CoMMpass and Mayo, was utilized to compare high CCND1 t(11;14) cases (≥300 transcripts-per-million [TPM], n=271) vs. low CCND1 hyperdiploid cases (<30 TPM, n=233). Two CCND1 knockout (KO) clones in t(11;14) MM U266B1 were also generated by CRISPR/Cas9, introducing a frameshift insertion (A) after chr11:69641478 (hg38) in exon 1. ATAC-seq and ChIP-seq (H3K4me3, H3K4me1, and H3K27ac) were performed in duplicates, and RNA-seq in triplicates in U266B1 wild-type (WT) and KO. Libraries were sequenced on Illumina NextSeq 2000 (2x51 bp for ChIP/ATAC, 2x150 bp for RNA-seq). Reads were mapped to hg38 by BWA (ChIP/ATAC) or STAR (RNA-seq), peaks called by MACS2 (FDR< 0.01), peak and gene raw counts estimated by featureCounts. Differential peaks (fold-change ≥1.5) and differentially expressed genes (fold-change ≥2) were identified by edgeR (FDR<0.05), including peaks detected in ≥2 samples and genes with TPM ≥0.1 in ≥1 sample. Interactions between promoters and distal regulatory regions were identified using publicly available chromatin interaction data in blood cell types. Pathway analysis was performed using Enrichr. Co-immunoprecipitation was performed in t(11;14) KMS12BM cells. Differential analysis comparing CCND1-high t(11;14) and CCND1-low MM cases in the two patient cohorts identified 21 upregulated and 21 downregulated genes associated with CCND1 expression which were also captured in the wild-typevs. CCND1 KO U266 cells. Enriched pathways (p<0.05) included Notch signaling, early/late estrogen response, epithelial-mesenchymal transition and reduced mitotic spindle and myogenesis. Dysregulated genes included those related to apoptosis, ubiquitin-proteosome, extracellular matrix, adhesion, and immune response, demonstrating a broader role for cyclin D1 in MM beyond cell cycle control. To explore the epigenomic basis of this transcriptional regulation, for each of the 42 differentially expressed genes we examined the changes of histone marks and chromatin accessibility in promoter and the interacting distal regulatory regions in the U266 KO model. Ten of the 21 genes upregulated in CCND1-high (downloaded in KO; e.g., argininosuccinate synthase 1 [ASS1], PMAIP1, MYEOV, CCR7, and DTX1) and 12 of the 21 genes downregulated in CCND1-high (upregulated in KO; e.g., UACA, RASAL2, PTPRZ1, PPT2, and IL23R) showed concordant changes in histone marks and/or chromatin accessibility, predominantly in distal regulatory regions, suggesting a chromatin-mediated mechanism of gene regulation by cyclin D1. The most prominent changes involved H3K27ac, a mark defining active enhancers. As H3K27ac is deposited by histone acetyltransferases CBP, we tested for the possible cyclin D1-CBP interaction. Co-IP confirmed this interaction, suggesting a role for this complex in chromatin remodeling and transcriptional control in t(11;14) MM. One gene notably upregulated in association with CCND1 is ASS1, which encodes a urea cycle enzyme responsible for endogenous arginine synthesis. Metabolomic studies demonstrated KO cells had elevated aspartate and citrulline and lowered arginine and ornithine compared to WT cells (p<0.01) demonstrating impaired urea cycle function. Further, KO cells displayed significantly reduced ASS1 protein expression, cell growth and S-phase entry compared to WT cells in arginine-replete conditions, with both WT and KO cell types further inhibited by arginine depletion (p<0.05). Together, a cyclin D1-associated gene signature was identified supporting its role in t(11;14) MM beyond the CDK4/6 mediated cell cycle control, including a role in metabolic plasticity and survival under arginine depleted conditions. These findings highlight cyclin D1 as a therapeutic vulnerability and open new avenues for personalized treatment strategies.

Abstract Motivation Efforts to address health disparities are often limited by the lack of robust computational tools for inferring genetic ancestry by calculating an individual’s genetic similarity to continental groups. We have already shown that a preferred alternative to self-described race is using ancestry-informative markers (AIMs) that can be classified into ancestral components and used to estimate their similarity to those of known populations to identify continental groups. However, real-world genomic data can present challenges, including limited availability of germline DNA, a small number of AIMs for each sample, and the use of different variant calling software, limiting the application of existing solutions. Results Here, we describe a novel supervised machine-learning tool AncestryGeni, which infers genetic ancestry for samples with even a hundred markers and is applicable to any genomic data, including whole exome sequencing (WES) and RNA sequencing (RNA-Seq) data. Applying AncestryGeni to a real-world genomic dataset obtained from the Multiple Myeloma Research Foundation (MMRF) CoMMpass study, we show that it is more accurate than the commonly used FastNGSadmix when using nonstandard genomic material. We also demonstrate that when using AncestryGeni, the tumor-derived sequence obtained from WES and RNA-Seq can be a robust data source to accurately estimate an individual’s genetic similarity to a continental group. Availability and implementation AncestryGeni pipeline is available at https://github.com/eelhaik/AncestryGeni/tree/main.

Introduction: Multiple myeloma (MM) with t(11;14)(q13;q32), resulting in CCND1;IGH rearrangement, remains the only subset amenable to targeted therapy with the BCL2 inhibitor, Venetoclax. Nevertheless, not all patients with this translocation respond to treatment and some progress early after initial response. The mechanisms behind this clinical heterogeneity are largely unknown. Methods: To investigate mechanisms of heterogeneity in response to Venetoclax, we interrogated 44 whole genome sequencing (WGS) and 24 whole exome sequencing data from a training cohort including 33 patients with t(11;14) MM treated with Venetoclax at the Mayo Clinic. In a validation cohort, we extracted single nucleotide variants (SNV) and indels from 17 MM patients with t(11;14) with available RNAseq data treated with Venetoclax at Emory University. Results: In the training cohort the median number of lines of therapy was 4 and Venetoclax drug regimens included combinations with dexamethasone, proteasome inhibitors, anti-CD38 agents, and immunomodulatory agents. After a median follow up of 30 months, the median progression free survival (PFS) was 9 months, and 13 patients (39%) achieved a complete remission (CR) to therapy, while 9 (27%) were refractory (i.e. no response or progressed within the first 60 days). For our driver discovery analysis, we integrated SNV, indels, copy number variants (CNV) and structural variants (SV). Deletion 17p (del17p) or TP53 mutation, MCL1 gain/amp, and hyper-APOBEC did not correlate with PFS in our cohort (p=0.59, p=0.71, and p=0.05, respectively). In contrast, mutations involving the MAPK pathway (NRAS, KRAS, BRAF, FGFR3) displayed the strongest association with short PFS, which was retained after adjusting for del17p/TP53 mutations, and 1q gain/amp (HR: 4.76, CI: 1.8-12.9, p=0.002). The association between suboptimal response and resistance to Venetoclax in MM patients with MAPK pathway mutations (n=7) was also observed in our validation cohort from Emory University (p=0.046). This is in line with what has been reported in functional studies in acute myeloid leukemia, where RAS pathway promotes MCL1 under the selective pressure of Venetoclax (Zhang et al. Signal Transduct Target Ther 2022). While MAPK mutations explained most cases of refractory patients (86%), the mechanisms through which initially responding patients subsequently progressed remain unclear. To investigate this, we interrogated 10 patients with pre- and post-treatment samples collected at disease progression (PD). We identified genomic events acquired/selected at PD of genes involved in the anti-apoptotic MCL1/BCL2 and MAPK pathways in 6/10 cases. One patient was refractory to Venetoclax and had double minutes involving BCL2, which caused multiple focal amplifications (>30 copies), likely reflecting extrachromosomal DNA. Another patient progressed with an acquired, focal deletion on BCL2L11. In two patients, the progressing clones harbored mutations in MAPK pathways (one BRAF and the other WNK2). Two patients who relapsed after sustained CR had major branching evolution where a previously undetectable focal deletion of PMAIP1 (NOXA) was seen at PD. Importantly, one of these deletions was a biallelic event. Venetoclax sensitivity assays conducted on PMAIP1 knockout in the MM t(11;14) cell line, KMS12PE, demonstrated significantly reduced cell death across all drug concentrations (p<0.05), validating the loss of NOXA as a mechanism of resistance to Venetoclax in MM. Finally, similarly with what was previously reported in chronic lymphocyte leukemia treated with Venetoclax (e.g., Herling et al Nat Comm 2018), CDKN2A/B monoallelic and biallelic focal deletions were selected post-Venetoclax, at PD in 4 patients while TP53 SNV/deletions were selected from subclonal to clonal in 2 patients at PD. Conclusion: In this study, we demonstrate that mutations involving the MAPK pathway identify a population of MM patients with t(11;14)(CCND1;IGH) who are unlikely to respond to Venetoclax. We also show that genomic events involving the MAPK and BCL2/MCL1 pathways are primarily responsible for disease progression after initial response. These findings highlight the critical role of comprehensive genomic profiling through WGS in identifying mechanisms of resistance to targeted therapies in MM, aiding patient selection for Venetoclax, and guiding the development of novel drug combinations.

Background: Multiple myeloma (MM) is an incurable plasma cell malignancy with several recurrent primary cytogenetic abnormalities including t(11,14)(q13;q32), resulting in increased expression of CCND1. Although 40% of patients with t(11;14) MM respond favorably to the BCL-2 inhibitor venetoclax (ven), 60% fail to achieve an objective response. Previous studies have suggested upregulation of PMAIP1, encoding the pro-apoptotic NOXA protein, in association with ven sensitivity (Leblay, Blood, 2024). Here we sought to evaluate the role for NOXA in ven response in t(11;14) MM. Methods: We analyzed whole exome, whole genome and RNA sequencing data of CD138+ tumor cells from 1,359 patients with MM in the CoMMpass and Mayo Clinic cohorts. The histone mark H3K27ac was assessed in t(11;14)+ KMS12PE and U266B1 human myeloma cell lines using ChIP-seq. Gene expression and protein levels of PMAIP1 (NOXA) were also detected by qRT-PCR and Western blot, respectively. Live cell detection was evaluated by Annexin V and PI using flow cytometry and by YoYo3 dye using IncuCyte following treatment with ven. Results: The t(11;14) MM subtype was associated with a 1.6-5.5-fold increased expression of PMAIP1 compared to non-t(11,14) MM in the CoMMpass and Mayo Clinic cohorts (p<0.0001). These findings are consistent with a significant correlation between CCND1 and PMAIP1 expression in both cohorts (p<0.001). To evaluate whether cyclin D1 is directly associated with increasing expression of PMAIP1 in t(11;14) MM, we generated a cyclin D1 knock out (KO) version of the U266B1 line using CRISPR/cas9. Two independent U266B1 KO cell lines showed reduced expression of PMAIP1 and NOXA protein compared to the WT U266B1 consistent with a role for cyclin D1 in regulating PMAIP1 expression. In addition, the KO lines had a 4.8-fold reduced H3K27ac signal at the PMAIP1 locus in comparison to the WT control suggesting a role for cyclin D1 in promoting H3K27ac. We previously observed increased expression of CCND1 in association with the G allele compared to the A allele of rs9344 in both the CoMMpass and Mayo Clinic cohorts (Leukemia, in press). We next asked whether the G risk allele of rs9344 within CCND1 impacted gene expression and H3k27ac levels at the PMAIP1 locus. To study the allele specific effect of CCND1, we used the CRISPR/cas9 edited KMS12PE WT (rs9344 AA) and Knock In (KI) (rs9344 GG) cell lines. In the KI cells, the G allele was accompanied by an ~2-fold increased expression of PMAIP1 and NOXA protein detected by western blot (<0.05). Increased PMAIP1 expression in KI lines was also accompanied by a 2.2-fold increased H3K27ac further suggesting a role for the G allele of rs9344 in promoting H3K27ac and gene expression at the PMAIP1 locus. Finally, we evaluated the consequence of PMAIP1 expression levels in relation to ven response. Increased expression of PMAIP1 and elevated levels of NOXA protein in the KMS12PE KI lines was associated with increased sensitivity to ven resulting in ~50% increase in cell death of KI vs. WT lines (p<0.02). Conclusion: We demonstrate a role for cyclin D1 in regulating PMAIP1 expression through H3K27ac resulting in increased NOXA protein. Levels of NOXA are associated with the G allele of rs9344 and increased venetoclax response.

Introduction: Cyclin D1, encoded by the CCND1 gene, is primarily known as a major regulator of cell cycle progression. Emerging studies have demonstrated a role for cyclin D1 in regulating gene transcription. In mantle cell lymphoma with t(11;14), cyclin D1 strongly binds to the promoters of highly transcribed genes and its overexpression led to genome-wide transcriptional downregulation. However, the transcriptional role for cyclin D1 in t(11;14) MM remains undefined. This study aims to elucidate the global chromatin and transcriptional changes induced by cyclin D1 in t(11;14) MM. Methods: Knockout (KO) of CCND1 in U266B1, a MM cell line carrying t(11;14), was performed by the Synthego (Redwood City, CA) using CRISPR/Cas9, which generated a frameshift insertion of base pair (bp) “A” after 69,641,478 bp within exon 1. Two independent KO clones were used for downstream analysis. ChIP-seq and ATAC-seq were performed for KO (n=2) and wild-type clones (WT, n=2). ChIP-seq used antibodies against H3K4me3 (promoter) and H3K4me1 and H3K27ac (enhancer). Libraries were sequenced to 2x51 bp on the Illumina NextSeq 2000 platform. Total RNA libraries (in triplicate) were sequenced to 2x150 bp. ChIP-seq and ATAC-seq reads were mapped to hg38 with BWA, and peaks were identified by MACS2 at FDR 0.01. RNA-seq reads were mapped to hg38 with STAR and gene raw counts were estimated using featureCounts. Differential peaks and differentially expressed genes between KO and WT were identified using edgeR at FDR 0.05 and fold-change 2. Only peaks present in at least 2 samples and genes having reads-per-million value 0.1 in at least 1 sample were included, excluding those from chrY. Pathway analysis was performed with the Enrichr package. Results: Differential expression analysis identified 2-fold more genes that were down-regulated than up-regulated in KO compared to WT (467 vs. 233). Pathway analysis revealed an enrichment of TNF-alpha signaling via NF-kappaB, inflammatory response, and apoptosis in association with down-regulated genes, while up-regulated genes are enriched in myogenesis, KRAS signaling, and epithelial mesenchymal transition pathway. In parallel, differential chromatin analysis identified 2.5% (1,161) of the peaks from H3K27ac and 0.5-1% (359-610) of the peaks for the other 3 marks that showed significant changes of signal levels between KO and WT. For the promoter mark H3K4me3, there are largely a comparable number of peaks showing increased (191 peaks) and decreased (168 peaks) signals, respectively, in KO compared to WT. However, for ATAC-seq and H3K4me1, about 70% of the differential peaks showed an increased signal in KO, in contrast with H3K27ac for which 60% of differential peaks showed decreased signal in KO. The preferential loss of H3K27ac occupancy is consistent with the observation of a higher proportion of genes being down-regulated in KO. These results indicated that knockout of CCND1 led to a global increase of chromatin accessibility, but a reduction of H3K27ac and gene expression. Further annotation revealed that, for H3K27ac and H3K4me1/ATAC-seq, 92% and 97% of the differential peaks were located >2kb away from the transcription start sites (TSSs) of expressed genes. Even for H3K4me3, 88% of the differential peaks are >2kb away from TSSs. The results suggest that the changes of chromatin upon CCND1 knockout predominantly occurred at distal regulatory regions. For up- and down-regulated genes, we estimated ChIP-seq and ATAC-seq signals in the TSS 2kb regions, revealing that up-regulated genes were overall associated with increased signals. Further analysis found that, for ATAC-seq and H3K4me3, 44% and 59% of the differential peaks located in the promoter regions were associated with differentially expressed genes, which dropped to ~20% for the two enhancer marks. These data indicate that changes in chromatin state impact the expression of the associated genes. Conclusion: Knockout of CCND1 impacts local chromatin state, particularly at enhancer regions, and the transcriptional program. In promoter-proximal regulatory regions, the changes of chromatin accessibility and H3K4me3 are strongly correlated with the expression changes of nearby genes. Further studies will identify the genes targeted by the differential enhancers and their possible roles in MM pathogenesis mediated by cyclin D1 following the t(11;14) event.